Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
1
Determination of migration of ion-implanted Ar and Zn in silica by
backscattering spectrometry
E. Szilágyia, I. Bányásza, E. Kótaia, A. Németha, C. Majorb, M. Friedb and
G. Battistigb
aInstitute for Particle and Nuclear Physics, Wigner Research Centre for Physics,
Budapest, Hungary
bInstitute for Technical Physics and Materials Science, Centre for Energy Research, H-
1121 Budapest, Konkoly Thege Miklós út 29-33, Hungary
Corresponding author: E. Szilágyi
Institute for Particle and Nuclear Physics, Wigner Research Centre for Physics,
H-1121 Budapest, Konkoly Thege Miklós út 29-33, Hungary
E-mail: [email protected]
Short biographical notes on all contributors.
Edit Szilágyi is the head of the Department of Materials Science by Nuclear Methods of the
Wigner Research Centre for Physics. She graduated in physics at Kossuth Lajos University,
Debrecen (1987), Ph. D. (Kossuth Lajos University, 1990), C.Sc. (1994) and D.Sc. (2011) of
the Hungarian Academy of Sciences. Her work includes nuclear analysis using MeV energy ion
beams (RBS - Rutherford Backscattering Spectrometry, ERDA - Elastic Recoil Detection
Analysis, NRA - Nuclear Reaction Analysis, channelling). She developed a computer code to
calculate the depth resolution of the above ion beam techniques. She has been extensively
dealing with ion implantation and with fundamental processes of ion implantation in porous
silicon, metals, and semiconductors.
István Bányász is a senior researcher at the Department of Materials Science by Nuclear
Methods of the Wigner Research Centre for Physics. He received his Ph.D. in Physics from
Loránd Eötvös University of Budapest, Hungary in 1987. He has worked in the fields of high-
resolution holography, holographic non-destructive testing and holographic recording materials.
His current research area is the application of ion- and electron beam irradiations to the
2
fabrication of optoelectronic components. Dr. Bányász is member of the Roland Eötvös
Physical Society and Senior Member of SPIE - the International Society for Optics and
Photonics.
Endre Kótai is senior research fellow of the Department of Materials Science by Nuclear
Methods of the Wigner Research Centre for Physics. He graduated in physics at Loránd Eötvös
University of Budapest. His main fields are Ion Beam Analysis (RBS, RBS/c, ERDA, NRA)
and ion implantation. He developed computer codes to simulate, evaluate IBA spectra.
Attila Németh is senior research fellow of the Department of Materials Science by Nuclear
Methods of the Wigner Research Centre for Physics. He graduated in physics at Uzhgorod State
University (1994). From 1994 to 2003 he worked as research assistant in Institute of Electron
Physics of National Academy of Scienses of Ukraine, Uzhgorod. In 2003 he obtained the title of
candidate of physical-mathematical sciences in field of physics of electron-ion collisions and
vacuum-ultraviolet spectroscopy (localized to Ph. D. in 2005). His present activities:
development of ion sources, development, optimization and application of ion implantation
techniques.
Csaba Major is currently working for the Department of Photonics in Res. Inst. for Technical
Physics and Materials Science, Budapest, Hungary. Main research fields: Optical system
designing, ellipsometer hardware development, evaluation of ellipsometric measurement.
Miklós Fried is currently the Head of Department of Photonics in Res. Inst. for Technical
Physics and Materials Science, Budapest, Hungary. His main research field is: Investigation of
surface modification of materials by ellipsometry and ion backscattering spectrometry.
Gábor Battistig, head of the Microtechnology Department, graduated in Electronic Technology
in Faculty of Electrical Engineering, Technical University of Budapest (BME) in 1981. From
2006 he is the head of the Microtechnology laboratory of MFA. His research interests include
various analytical methods for materials characterizations especially ion beam techniques. He is
working in the development of Si based 3D MEMS devices as integrated gas, bio, chemical and
mechanical sensors and the related processing technology with special emphasis on ion
implantation and thin layer depositions.
3
Determination of migration of ion-implanted Ar and Zn in silica by
backscattering spectrometry
It is well known, that the refractive indices of lots of materials can be modified
by ion implantation, which is important for waveguide fabrication. In this work
the effect of Ar and Zn ion implantation on silica layers was investigated by
Rutherford Backscattering Spectrometry (RBS) and Spectroscopic Ellipsometry
(SE). Silica layers produced by chemical vapour deposition technique on single
crystal silicon wafers were implanted by Ar and Zn ions with a fluence of
1-21016 Ar/cm2 and 2.51016 Zn/cm2, respectively. The refractive indices of the
implanted silica layers before and after annealing at 300 and 600 oC were
determined by SE. The migration of the implanted element was studied by real-
time RBS up to 500 oC. It was found that the implanted Ar escapes from the
sample at 300 oC. Although the refractive indices of the Ar implanted silica
layers were increased compared to the as-grown samples, but after the annealing
this increase in the refractive indices is vanished. In case of the Zn implanted
silica layer both the distribution of the Zn and the change in the refractive indices
were found to be stable. Zn implantation seems to be an ideal choice for
producing waveguides.
Keywords: waveguides, ion implantation, silica, chemical vapour deposition,
Rutherford backscattering spectrometry, spectroscopic ellipsometry
Introduction
Optical waveguides have become basic elements of integrated optics and
optoelectronics. Planar and channel waveguides are characterized by refractive indices.
Several techniques are developed to set up and adjust refractive indices of optical
materials (glasses, polymers, crystals, etc.). Further microelectronic technology can be
used to fabricate integrated optics; therefore the thermal stability of waveguides is also
important. Ion implantation, compared with other waveguide fabrication methods, has
some unique advantages. It has proved to be a universal technique for producing
waveguides in most optical materials. Moreover, ion implantation provides better
4
controllability and reproducibility in comparison with other techniques, e.g. diffusion of
metal ions, ion exchange, sol gel and layer deposition techniques (for example
sputtering, molecular beam epitaxy, and pulsed laser deposition, chemical vapour
deposition). The first ion implanted waveguides were produced in 1968 by proton
implantation into fused silica glass (1), and the changes in refractive indices due to H+,
He+ and N+ ion implantations have been characterized by several other groups (2, 3, 4,
5).
SiO2 is an important insulating material with broad applications in the
semiconductor and glass industries; therefore the thermal stability of the modified ion
implanted layers is always an important question during device fabrication. In silica, the
diffusion coefficient of several elements is higher compared to other materials (6); it is
extremely high for example for He. Helium accumulation, migration and diffusion are
intensively studied in nuclear materials because it has a strong influence on
microstructural, physical and thermo-mechanical properties of natural or manufactured
inorganic solids (7, 8). Because of the extremely high diffusion coefficient of the ion-
implanted helium in silica, helium is able to escape from SiO2 films, even from
substoichiometric SiOx films, if x > 1.3 (9). The helium can only be forced in silica, if
buried silica island is formed (10), where the oxide is insulated properly from the
vacuum.
The thermal stability of a layer, diffusion or migration of a given element can be
studied by in-situ annealing combined with Rutherford Backscattering Spectrometry
(RBS). The in-situ annealing/RBS technique is already mentioned in the proceedings of
the 1st Ion Beam Conference (11), however, recently referred as real-time RBS is
frequently used by several groups to study solid state reactions (12, 13).
5
In this work the effect of Ar and Zn ion implantation into CVD oxide layer were
investigated in order to determine whether large enough increases in the refractive
indices, suitable for waveguides applications, are produced. Ar is a noble gas similar to
He, no chemical reaction with the target atoms occurs, while Zn more probably forms
bonds with either Si or O. Mazzoldi and his co-workers published a comprehensive
review on peculiarities and application perspectives of metal ion implantation in glasses
in 1994 (14). They presented in detail experimental evidences of chemical interactions
between the implanted metal ions and the target atoms, including Ti and W in pure SiO2
and in silica glasses. Formation of nanometer-sized colloidal particles after
implantation with metal ions was also demonstrated. Gao et al. reported synthesis of
ZnS nanocrystallites in SiO2 via multiple ion implantation and rapid thermal annealing
(15). Amekura and his co-workers produced cupric oxide nanoparticles in silica glass
by implantation of 60 keV Cu- ions and subsequent thermal annealing in oxygen
atmosphere (16). On basis of the above mentioned results, formation of layers with
high concentration of ZnO nanoparticles (with a considerable refractive index change)
in the SiO2 thin layer could be expected.
Spectroscopic Ellipsometry (SE) and RBS were used to study the changes in
refractive indices, and to determine the elemental composition of implanted films. Ar
and Zn migration were followed by in-situ annealing/RBS technique. Since both Ar and
Zn are heavier than the elements of the substrates, RBS is an ideal method to study their
migration in the target material.
Experiment
In the present experiment first an oxide layer with a nominal thickness of 110 nm was
formed on single crystal silicon (100) substrate by atmospheric pressure chemical
6
vapour deposition (CVD) technique. The deposition temperature was 450 oC. The whole
wafer was cut to several pieces for implantation.
Two sets of three samples (each) were implanted by 50 keV Ar with a fluence of
11016 at/cm2 or 21016 at/cm2; two of them were implanted by 130 keV Zn ions with a
fluence of 2.51016 at/cm2. One of the Ar implanted samples was annealed for 1 h at
300 oC, while the other one at 600 oC. The third samples are referred as “as-implanted”.
All the implanted and annealed samples as well as one of non-implanted
samples were characterized by SE and RBS.
The optical properties and the thicknesses of thin film structures can be derived
from (Ψ,Δ) values measured by SE, where Ψ and Δ describe the relative amplitude and
relative phase change of polarized light during reflection, respectively. In the present
experiment Ψ and Δ were measured by a Woollam M-2000DI rotating compensator
ellipsometer in the 300 – 1700 nm wavelength range at angle of incidence of 70o and
75o. The refractive index of the layers was described by the Cauchy dispersion relation
(17), an empirical polynomial approach for refractive index function of insulators and
semiconductors below the band-gap (18). The calculated (generated) spectra were fitted
to the measured ones using a regression algorithm. The measure of the fit quality is the
mean square error (MSE) which was compared for different optical models. The
unknown parameters are allowed to vary until the minimum of MSE is reached.
For standard RBS, the samples were fastened to a simple sample holder of a
scattering chamber equipped with a two-axis goniometer. During the experiments the
vacuum in the chamber was better than 1 10-4 Pa using liquid N2 traps along the beam
path and around the sample. The ion beam of 2000 keV 4He+ was collimated with 2 sets
of four-sector slits to the necessary dimensions of 0.5 0.5 mm2. The ion current of
typically 10 nA was kept constant via monitoring by a transmission Faraday cup (19).
7
The spectra were collected with a measurement dose of 4 μC. The RBS measurements
were performed with an ORTEC surface barrier detector under a solid angle of
4.14 msr. The energy calibration of the multichannel analyser was performed by using
known peaks and surface edges of Au, Si and C, respectively. To identify the surface
elements and buried peaks, the RBS experiments were performed at least at two
different tilt angles.
The hydrogen content of the as prepared oxide layer was determined by Elastic
Recoil Detection Analysis (ERDA) using reflexion geometry at tilt 80o and recoil angle
of 20o using 1600 keV He beam. The solid angle of the ERDA detector was 0.8 msr. To
stop the forward scattered 4He+ ions a Mylar foil of 6 m thickness was applied before
the ERD detector. To determine the hydrogen loss caused by the beam itself, 20
independent measurements were carried out with a dose of 0.2 C each (therefore total
dose was 4 C).
To study the thermal stability of the layers in-situ annealing combined with RBS
was carried out using a heated sample-holder. The temperature increased from room
temperature (RT) to 500 oC with a linear ramping rate of 2.3 oC/min. In the case of Zn
implanted sample 500 oC was kept for1 h. During the experiment RBS spectra were
continuously collected with a dose of 1 C using 2000 keV He+ beam. The collection
time of a spectrum was about 40 s, i.e., hundreds of spectra were recorded. To improve
the statistics several spectra can be summed during the evaluation, if it is necessary.
All the spectra taken on a sample were than simulated using the same layer
structure obtained from the RBX simulation program (20).
Results and discussion
In the case of as prepared CVD oxide sample the elemental composition found to be
8
SiO2.4H0.16. The oxygen excess in the CVD oxide can be explained by oxygen excess-
related defects (21). From the thicknesses given by ellipsometry (dSE) and RBS (dRBS, in
at/cm2 units) the density of the layer can be easily calculated as =dRBS/dSE, resulting
6.181022 at/cm3.
Ar implanted samples
The samples were implanted with two different Ar fluences. A maximum concentration
of 0.87 at% and amount of ~81015 Ar/cm2 has been found in the sample independently
on the implanted fluences. RBS spectra of Ar implanted samples with fluences of
11016 at/cm2 and 21016 at/cm2 are shown in Fig. 1. Their depth distribution is far from
the profile calculated by SRIM (22), Ar extends to the whole oxide layer. These results
can be interpreted assuming limited number of traps for argon in the oxide matrix; if all
the traps are occupied, the excess Ar can freely move in the layer and it can escape from
the layer. Comparable results are found in similar fluence range for fused silica and
soda-lime glass samples (23). These facts underline that this is an intrinsic property of
silica, controlled by diffusion coefficients and the solubility limits, and it does not
depend on the preparation methods.
Applying 1 h annealing in the implanted sample at 300 or 600 oC, all the Ar
leaves the oxide layers. RBS shows that the composition and the layer thickness slightly
changed, however, from these experiments we could not decide whether it is a real
effect or only exhibits some inhomogeneity of the samples. As shown in Fig. 2, the
refractive indices as a function of the incident wavelength determined by SE also show
only small changes for the two different fluences. In case of the annealed samples, as
the implanted ions escaped, the changes in the refractive indices are caused by only the
structural modifications. The highest refractive indices were found for the case of the
9
as-implanted samples, then the refractive indices decrease as a function of temperature.
After 600 oC annealing the refractive indices became lower than the as prepared sample.
The densities of all Ar implanted oxide layer increase up to 6.6–6.81022 at/cm3, that
shows densification of the CVD oxide. These values are close to the density of thermal
oxide (6.671022 at/cm3).
Fig. 3 shows the results of in-situ annealing combined with RBS experiment
performed on the sample implanted by low Ar fluence. The Ar peak area was
determined as a function of the temperature. The uncertainty of the Ar peak areas was
calculated on the basis of the uncertainty budget of the experiment (24). To increase the
statistics, two successive spectra were summed. The Ar starts to escape at 225 oC, and
there is another significant step at 350 oC. Finally, Fig. 4 presents the comparison of the
spectra taken at RT and 450 oC, clearly showing the changes both in the composition
and the layer thicknesses: not only the Ar escapes, but some oxygen loss is also
occurred as the composition changed from SiO2.3Ar0.03 to SiO2.1. The amount of silicon
atoms in the oxide layer remains constant within experimental uncertainty.
Zn implanted samples
In case of Zn implanted sample the in-situ annealing/RBS measurement was carried out
using the same linear ramping as it was used in the previous experiment. After reaching
500 oC, this temperature was kept for 1 h. The experiment proves that no change in the
composition can be observed, as shown in Fig. 5.
The RBS profile shows that the Zn mainly stopped in the silica layer, but some
particles reached the interface and stopped in the silicon. SE results can be interpreted,
therefore, by a three-layer model. Two of them are the upper (low Zn-concentration)
and the bottom (high Zn-concentration) part of the oxide layer with very different
10
behaviour of refractive indices as a function of wavelength, the third one is a thin
partially amorphized layer on the substrate, next to the oxide. Only small changes in the
refractive indices of oxide layer are observed due to 1h annealing at 500 oC, as shown in
Fig. 6. Let us note that Cauchy dispersion relation is an empirical polynomial approach
for refractive index function of insulators and semiconductors below the band-gap, so
the reason of the different sign of dispersion below 500nm wavelength (see Fig. 6.c) is
unknown. Maybe different phases (ZnO nanoparticles, mixed SiZnO phases) are
involved. However, the wavelength region above 600 nm is the relevant region for the
case of the waveguide applications. In spite of the fact that the implanted Zn ions
entered into the Si substrate resulting partially amorphisation on the Si surface, the
amplitude of the refractive index modulation across the SiO2 layer reaches about 0.2 in
the visible range. So fabrication of waveguide structures by this method seems to be
feasible.
There are two significant differences between the Ar and Zn implanted samples:
while Ar implantation changes the refractive indices of the whole layer, Zn modifies
only a part of the layer. Moreover Zn implantation results much higher increase in the
refractive indices than Ar implantation, and the increase is stable up to 500 oC.
Conclusions
The deposited CVD oxide was found to be not stoichiometric. A composition of
SiO2.4H0.16 was measured with a lower density compared to the thermal oxide.
In case of the Ar implantation 0.87 at.% Ar is found in the oxide layer
independently of the implanted fluence. Ar is not located near to the range expected
from the SRIM calculation but extends to the whole oxide layer. The density of the
layer increased. During 1 h ex-situ annealing at 300 oC, all the Ar leaves the layer.
11
Changes in refractive indices therefore can be interpreted by structural changes due to
Ar implantation and annealing.
In-situ RBS results show that at least two Ar states exist in SiO2, lower energy
states and higher energy states release Ar at 225 oC and above 350 oC, respectively. The
composition changed from SiO2.4Ar0.03 to SiO2.1.
In case of Zn implantation no change in the composition can be observed during
thermal annealing between RT and 500 oC. Ellipsometry results can be interpreted by
three layer model. Two of them connected to the oxide layer with very different values
of refractive indices as a function of wavelength, the third one is a thin partially
amorphized Si layer on the substrate. Zn implantation causes much higher increase in
the refractive indices than Ar implantation, and this increase is stable up to 500 oC.
Therefore Zn implantation seems to be useful for waveguide applications.
Acknowledgement
The partial support of the Hungarian Scientific Research Fund (OTKA Grants No.
K101223 and K112114) is acknowledged. Z. Zwickl is gratefully thanked for his
technical help during the experiments.
(1) Schineller E.R.; Flam R.P.; Wilmot D.W. J. Opt. Soc. Am. 1968, 58, 1171–1173.
(2) Townsend P.D.; Chandler P.J.; Zhang L Optical effects of ion implantation; Cambridge
University Press: Cambridge, UK, 1994.
(3) Peña-Rodríguez O.; Olivares J.; Carrascosa M.; García-Cabañes Á.; Rivera A.; Agulló-
López F. (2012). Optical Waveguides Fabricated by Ion Implantation/Irradiation: A Review.
In Ion Implantation, Goorsky M. Prof. Ed., ISBN: 978-953-51-0634-0, InTech, 2012;
Chapter 12.
References
12
(4) Bányász I; Berneschi S.; Bettinelli M.; Brenci M.; Fried M.; Khanh N.Q.; Lohner T.; Nunzi
Conti G.; Pelli S.; Petrik P.; Righini G.C.; Speghini A.; Watterich A.; Zolnai Z. IEEE
Photonics J., 2012, 4, 721–727.
(5)Bányász I.; Zolnai Z.; Fried M.; Berneschi S.; Pelli S.; Nunzi Conti G. Nucl. Instr. Meth. B,
2014, 326, 81–85.
(6) Francois-Saint-Cyr H.G.; Stevie F.A.; McKinley J.M.; Elshot K.; Chow L.; Richardson K.A.
J. Appl. Phys. 2003, 94, 7433–7439
(7) Trocellier P.; Miro S.; Serruys Y.; Vaubaillon S.; Pellegrino S.; Agarwal S.; Moll S.; Beck
L. Nucl. Instr: Meth. B 2014, 331, 55–64.
(8) P. Trocellier P.; Agarwal S.; Miro S. J. Nucl. Mater., 2014, 445, 128–142.
(9) Manuaba A.; Pászti F.; Ramos A.R.; Khánh N.Q.; Pécz B.; Zolnai Z.; Tunyogi A.; Szilágyi
E. Nucl. Instr: Meth. B. 2006, 249, 150–152.
(10) Szakács G.; Szilágyi E.; Pászti F.; Kótai E. Nucl. Instr: Meth. B. 2008, 266, 1382–1385.
(11) Baglin J.E.E.; Hammer W.N. in: Meyer O.; Linker G.; Käpeller F. (Eds.), Ion Beam
Surface Layer Analysis, Plenum Press, New York, 1976, p. 447.
(12) Theron C.C.; Lombaard J.C.; Pretorius R. Nucl. Instr: Meth. B. 2000, 161–163, 48–55.
(13) Demeulemeester J.; Smeets D.; Van Bockstael C.; Detavernier C.; Comrie C.M.; Barradas
N.P.; Vieira A.; Vantomme A. Appl. Phys. Lett. 2008, 93, 261912.
(14) Mazzoldi P. ; Arnold G.W. ; Battaglin G. ; Bertoncello D.; Gonella F. Nucl. Instr: Meth.
B. 1994, 91, 478–492.
(15) Gao K.Y.; Karl H.; Grosshans I.; Hipp W. ; Stritzker B. Nucl. Instr: Meth. B. 2002, 196,
68–74.
(16) Amekura H.; Kono K; Takeda Y.; Kishimoto N. Appl. Phys. Lett. 2005, 87, 153105.
(17) Langhoff P.; M. Karplus M. J. Opt. Soc. Am. 1969, 59, 863–871
(18) Jenkins F.; White H. Fundamentals of Optics; (3rd ed.) McGraw-Hill, New York, 1957
(19) Pászti F.; Manuaba A.; Hajdu C.; Melo A.A.; da Silva M.F. Nucl. Instr: Meth. B. 1990, 47
187–192.
(20) Kótai E. Nucl. Instr: Meth. B. 1994, 85, 588–596.
13
(21) Skuja L.; Kajihara K.; Hirano M.; Hosono H. Nucl. Instr: Meth. B. 2012, 286, 159–168.
(22)Ziegler J.F.; Ziegler M.D.; Biersack J.P. Nucl. Instr. Meth. B. 2010, 268, 1818–1823
(23) Battaglin G.; Arnold G.W.; Mattei G.; Mazzoldi P.; Dran J.-C. J. Appl. Phys. 1999, 85,
8040–8049.
(24) Jeynes C.; Barradas N.P.; Szilágyi E., Anal. Chem. 2012, 84, 6061–6069.
14
Figures
Figure1. RBS spectra taken on Ar implanted samples with fluence of 11016 at/cm2 (L0)
and 21016 at/cm2 (H0). Inset shows the concentration profiles of Ar, and for the sake of
the comparison the SRIM simulation is also shown that normalized the same amount of
Ar.
15
Figure 2. Refractive indices for Ar implanted samples. For comparison, the refractive
indices of virgin sample and the thermal SiO2 (from literature) are also shown.
Figure 3. In-situ annealing combined with RBS experiment performed on the Ar
implanted samples with fluence of 11016 at/cm2.
16
Figure 4. Composition changes. RBS spectra taken at RT and 450 oC during the in-situ
annealing experiment. To increase the statistic 10 successive spectra were summed, that
correspond to a dose of 10 C.
Figure 5. RBS spectra taken on Zn implanted sample at RT and 500 oC after 1 h
annealing. Tilt was 30o; the dose of the spectra was 10 C.
17
18
Figure 6. a) SE results on the Zn implanted sample after 1 h annealing at 500 oC. b)
Comparison of depth profiles evaluated from RBS and SE. c) refractive indices as a
function of wavelength for the two sublayers. For the sake of comparison the refractive
indices of the virgin oxide as well as the thermal oxide are also shown.